PENAPH

Participatory Epidemiology and the Use of Models to Design Control Strategies

c/o ILRI, P. O. Box 30709, Nairobi, 00100 Kenya; phone: +254-20 422 3000; fax:+ 254-20 422 3001; email:ilri-kenya@cgiar.org

Participatory Epidemiology

The use of participatory rural appraisal techniques to collect epidemiological

knowledge and intelligence

Participatory Rural Appraisal (PRA)

• Qualitative intelligence gathering process

• Key informants• Problem solving

– Multiple methods– Multiple perspectives– Triangulation

• Best-bet scenarios

Existing Veterinary Knowledge

• Traditional terms and case definitions

• Clinical presentation• Pathology• Vectors• Reservoirs• Epidemiologic

features Photo: T. Leyland

Applications

• Basic Research • Active Surveillance

– Participatory Disease Surveillance (PDS)• Holistic Needs Assessment

– Stakeholders, livelihoods and risk• Impact Assessment

– Participatory Impact Assessment (PIA)– Qualitative and Quantitative

• Institutional change

Participatory Disease SurveillancePDS

• Active surveillance done by professionals

• Risk-targeted• High detection rate

– Information networks– Extended time frame

• Sensitive and Specific– Validation processes– Laboratory support

• Timely

Attributes of PE/PDS Programs

• Flexible approach that allows for discovery• Practitioners are problem-solvers and not enumerators• Strength of the approach lies in its flexible and qualitative

nature• Orients and complements, but does not replace structured

and quantitative methods• Information from diverse sources and methods• Analyzed in an iterative process referred to as

triangulation• Integrates biological testing and quantitative methods

when appropriate to objectives

PENAPHParticipatory Epidemiology Network for Animal and

Public Health

c/o ILRI, P. O. Box 30709, Nairobi, 00100 Kenya; phone: +254-20 422 3000; fax:+ 254-20 422 3001; email:ilri-kenya@cgiar.org

• Nine core partners

•Building Surveillance Capacity

• Good Practice Guidelines

• Certification of Training

• Research, Policy and Advocacy

• Pro-Poor and One Health Focus

• Knowledge Exchange

Appropriate Combinations of Complimentary Techniques

• Participatory methods• Biological testing• Analytical methods

Persistence as a Function of Initial Herd Immunity

0

100

200

300

400

500

600

700

800

0 20000 40000 60000 80000 100000 120000 140000 160000 180000 200000

Initial Number Recovered (Immune)

Le

ng

th o

f O

utb

reak

(D

ays

)

Participatory Approaches to the Mathematical Modelling of

Rinderpest and CBPP

PARC, PACE and CAPE

Modelling Objective

• Model agents and lineages as they occurred in pastoral areas of East Africa

• Involve stakeholders at all levels in design and address principal questions of pastoralists and policy-makers– More useful and creates a sense of ownership

• Inform policy dialogue leading to more effective strategies– Targeting interventions and surveillance

Data Collection• Evidence-based literature review• Participatory epidemiology – expert opinion

– Mortality, prevalence, clinical course, spatial and temporal patterns and contact structure, estimation of R0

– SSI, proportional piling, matrix scoring, relative prevalence scoring, mapping and event trees

• Serology - Estimation of R0

“Some of us believe we have rinderpest, but we are not really sure. The disease looks like rinderpest, but it doesn’t kill the

animals. It is rinderpest-like or mild rinderpest.”

Somali elder on lineage 2 rinderpest

El Wak, Somalia 1996

Modelling Approach

• State-Transition• Stochastic

– Critical Community Size– Fade - Out

• Open Population• @ Risk Software• Input Parameters – Beta Pert Distributions• Single population and multipopulation• Outputs – Probability Distributions

E I RSb

μ μ μσμ

β γ α

α = recovery rateb = birth rateβ = effective contact rateγ = latency to infectious rateμ = non-specific mortality rateσ = RP mortality rate

S = SusceptibleE = ExposedI = InfectiousR = Resistant

RP Model Structure

ωr

ρ ωv

E I RSb

μ μ μσμ

β γ αr

Q

μ

V

αq

κψ

μ CBPP Model

The Basic Reproductive Number Ro

• Number of secondary cases resulting from one infected animal in susceptible population

• Ro is a feature of both the strain of infectious agent and the host population

The Importance of R0

• A measure of the transmission rate • Can be easily estimated from field data

– RP - Herd immunity threshold (1-1/R0)– CBPP - Average of first infection– HPAI – Final fraction size

• Effective contact rate, β, can be calculated from R nought.

• Herd immunity targets

Inter-Epidemic PeriodRinderpest in Sudan and Somalia

Temporal distribution of reports on rinderpest compatible events by year and interview area

So What?• Prevalence of Infection

– 0.1 – 0.2 %– Random clinical surveillance to OIE

standards (1%) not useful– Participatory disease surveillance

• Critical Community Size– ~200,000 head– Target vaccination to high risk populations– Large, remote pastoral communities

• Lineage 2 in Somalia– Modest herd immunity levels could

eradicate

Disease Persistence as a Function of the Initial Prevalence of Immunity

0

100

200

300

400

500

600

700

800

0 20 40 60 80 100

Initial Prevalence of Immune (%)

Days

Murle

Dinka

Toposa

Jikany Nuer

Lou Nuer

Jie

Niagatom

Gawere Nuer

Bor Dinka

Lotuko/Lopit

Laak Nuer

Boya/Didinga

Anuak

Kachipo

ProvincesRiversCattle Populations

300 0 300 600 Kilometers

N

EW

S

East Nile Cattle Populations

Traditional Livestock Exchange

Kachipo

1

2

3

Murle

Toposa

Jie

Heterogeneous RP Model for lineage 2 in the Somali Ecosystem

• Four populations• Multiple species

– Parameter values can set independently

• Eight year duration• Cattle and buffaloes• Buffalo R0 ~ 8

Four Sub-Population Epidemic Curves from a Heterogeneous Population Model for Rinderpest in Cattle Where the Between

Population Contact Rate is 1% of the Within Population Contact Rate

0

20

40

60

80

100

120

140

0 100 200 300 400 500 600 700

Days

Num

ber I

nfec

tious

Sub-population 1Sub-population 2Sub-population 3Sub-population 4

Results Over 2 Years

• Mean mortality 0.85% per year• 34% of iterations ended with a prevalence

of infection of 0.1 to 1%• Maximum prevalence 2%• Final average immunity 26%

Eight Year Model η = 0.005 and R = 35%

DurationMean 95th % Max Persistence

(%)

10,000 137 323 603 0

20,000 227 532 1100 0

30,000 329 878 1571 0

40,000 408 948 1915 0

50,000 477 1418 2633 0

100,000 1031 2553 2920 3

The Role of Buffaloes

Buffaloes % Immune

Cattle - Day of Last Case

Overall – Day of Last Case

5-15 245 247

15-25 264 269

25-35 261 266

45-55 258 265

55-65 253 260

Somali Ecosystem

• Buffalo act as an indicator host and do not contribute to persistence in cattle– Explosive outbreaks of short duration

• Small isolated communities can maintain Lineage 2 for prolonged periods

• Intensive surveillance and time

Potential of a Combined Vaccination and Treatment Strategy for CBPP

Scenario Baseline 1/α 75% 1/α 50% 1/α

None 75.4 178 59.6 129 33.2 64

Annual – 5 yr 67.8 130 43.0 79 7.6 24

Biannual – 2 yr 57.2 129 - - 4.4 20

Biannual – 5 yr 35.2 83 8.0 43 0.4 16

Indications from the CBPP Modelling

• Eradication not possible with existing vaccine without severe movement control

• The potential impact of treatment is at least great as available vaccines– Private sector and consumer driven application

• Research on treatment regimes merits the same level of attention and investment as vaccine development.

Conclusion

• Simple, intuitive models serve as good communications tools for underlying concepts

• Bring diverse information together to be tested as for biological coherence

• Choose between strategy options or identify new options

• Involve beneficiaries and decision-makers from the outset.